Eur@. J. Cancer Vol. 11, pp. 783-786. Pergamon Press 1975. Printed in Great Britain
Carcinoembryonic Antigen (CEA)and Related Antigens in Blood Cells and Hematopoietic Tissues* M. BORDES, S. KNOBEL and F. MARTIN Laboratory of Experimental iVIedicine, I N S E R M U-45, Facultd de Mbdecine, D~on, (France). Abstract--Indirect immunofluorescence was used to detect and localize carcinoembryonic antigen (CEA) and non-specific cross-reacting antigen ( NCA) in blood cells and hematopoietic tissues. CEA was neverfound, but NCA could easily be identified in the cytoplasm of blood polymorphonuclear cells, bone marrow myelo~ytes, and monocytic cells (probably macrophages) infiltrating the spleen and lymph nodes. In normal colon and colonic adenocarcinoma, CEA was present in the walls of epithelial cells, and in round, isolated polymorphonuclear or mononuclear cells, infiltrating the submucosa and the stroma of the tumor.
INTRODUCTION
MATERIAL AND METHODS
CARCINOEMBRYONIC antigen (CEA) was first defined as a cancer- and organ-specific antigen, restricted to cancerous and fetal digestive tissues [1]. This specificity was later questioned, and CEA was detected in various epithelial tissues and secretions [2, 3]. CEA, or an immunologically related antigen, has recently been demonstrated on the walls of normal erythrocytes [4]. A glycoprotein, sharing a common antigenic site with CEA, was independently substantiated by several groups, and called non-specific cross-reacting antigen(NCA) [5], normal glycoprotein (NGP) [6] and C E X [7]. Exchange of reagents has proven the immunologic identity of NCA, N G P and C E X [7]. This antigen has been found, in association with CEA, in normal and cancerous epithelial tissues [8], but was also present in normal spleen [5], an organ which does not contain epithelial tissue. The present work takes into consideration the location of CEA and NCA in blood cells and hematopoietic tissues.
Antigens and antisera Purified CEA was prepared in this laboratory, according to previously reported procedure [9]. Anti-CEA serum was prepared by repeated immunization of rabbit with purified CEA. This serum, referred to as 42/f, has both antiCEA and anti-NCA reactivities. It was made specific for CEA by absorption of perchloric acid extract from normal human spleen (50 mg/ml). This absorbed serum will be referred to as 42/f2. Blood and tissue specimens Samples of blood have been obtained from 20 healthy donors of various blood group status. Morphologically normal, sternal bone marrow samplings came from 5 patients with mild anaemia, of non-malignant origin. Smears of blood and bone marrow were immediately fixed for 20 min in 95 % ethanol. The other tissues were fixed and embedded according to Sainte-Marie [10]. The non-cancerous specimens were 2 spleens, surgically removed after a traumatic rupture, 2 cervical lymph nodes (biopsies), with mild subacute lymphadenitis,
Accepted 1 April 1975. *This work was supported by research grants CA-401 and 74.1.080.04 from the Institut National pour la Sant6 et la Recherche Mddicale. 783
Fig. 5. Colonic adenocarcinoma ~tained by anti-NGP antiserum (indirect immunofluorescence). The membrane of tumor cells (bottom, right), but also round cells infiltrating the stroma (arrow) are stained. ( x 550). Fig. 6.
Circulating blood polymorphonuclear cell stained by anti-NGP antiserum (indirect immunofluorescence). ( x 550).
Fig. 7.
A clone of cells of the granulopoietic line stained by anti-NGP antiserum (indirect immunoJtuorescence) on a bone marrow smear. ( x 550).
Fig. 8. Section of normal spleen, stained by anti-NGP antiserum. The lymphocytes of the germinal center are negative, but the round cells (polymorphonuclear and~or macrophages) around the center are stained. ( x 200).
Fig. 1. Reaction of anti-CEA antiserum (42/f2, central well) with purified CEA and perchloric acid extract of colonic carcinoma (CT) ; the serum does not react with perchloric acid extract of normal spleen (Sp). Fig. 2. illembrane immunofluorescence of blood group A erythrocytes with antiCEA serum, before absorption with A red blood cells. This fluorescence is completely extinguished by absorption of the serum with blood group A erythrocytes. ( x 550). Fig. 3. Reaction of anti-CEX antiserum (X, central well) with perchloric acid extracts of normal spleen (Sp), and polymorphonuclear cells (Pm) and purified NGP. Fig. 4. Reaction of an anti-CEA antiserum non absorbed for NCA (42f, central well) with purified CEA, perchloric acid extracts of colonic carcinoma (CT) and normal spleen (Sp).
(to face p. 784)
784
M. Bordes, S. Knobel and F. Martin
and a morphologically normal lung obtained at autopsy. Three colonic adenocarcinomas, and their adjacent, non-cancerous mucosa (surgical material), were also used in this work. Peritoneal macrophages These were prepared from the ascitic fluid of a patient with alcoholic liver cirrhosis. Five ml were disseminated into tissue culture flasks (Falcon), poured off after 2 hr, and replaced by Ham's FIO medium, supplemented by 20 % fetal calf serum. The adhering cells were mostly macrophages, as proven by phagocytosis of iron carbonyl. After 3 days of culture, they were fixed by absolute ethanol and tested for immunofluorescence. Immunofluorescence staining The indirect immunofluorescence technique, used to stain smears and sections, has been previously reported [11]. 42/f, 42/f2, anti-CEX, anti-NGP and anti-NCA sera were respectively diluted as follows 1/80, 1/80, 1/20, 1/20 and 1/100 according to preliminary results. In the second step, the fluorescein isothiocyanateconjugated reagents were anti-goat globulin rabbit antiserum (Institut Pasteur, Paris) diluted 1/10 for anti-CEX serum and antirabbit globulin sheep antiserum (Institut Pasteur, Paris) diluted 1/10, for the other sera. Special precautions were taken to avoid nonspecific reactions. All the smears and sections were treated by sodium nitrite-acetic acid in accordance with Burtin and Sabine [12]. Fluorescein labeled sera were absorbed by acetonic extracts of normal human spleen (150 mg/ml). Normal rabbit serum was systematically used as a negative control. Observations were made with a Leitz Orthoplan microscope with a vertical illuminator. Membrane immunofluorescence Erythrocytes from blood collected on sodium oxalate were washed 3 times in phosphatebuffered saline (PBS), incubated for 20 rain at room temperature with an equal volume of antiserum, diluted as mentioned above. The red blood cells were washed 3 times in PBS, then incubated in the same volume of fluorescein-labelled serum diluted 1/10. After 3 further washings, they were mounted on slides and examined for membrane immunofiuorescence Immunodiffusion Double immunodiffusion was carried out according to a modification of Ouchterlony's technique [2].
RESULTS
Normal rabbit serum On immunofluorescence, normal rabbit serum, used as control, does not stain ethanolfixed smears or tissue sections of hematopoietic tissues or colonic adenocarcinoma. Anti-CEA serum On immunodiffusion (Fig. 1), anti-CEA serum 42/f 2 gives a precipitin line with perchloric extract of colonic tumor or purified CEA, but does not react with perchloric extracts of spleen. The immunofluorescence staining of colonic tumors and non cancerous colonic mucosa by this serum has been described elsewhere [11]. When tested on fixed blood or bone marrow smears, serum 42/f2 stains the membrane of erythrocytes and erythroblasts from patients of blood group A or AB, but not of O or B. Membrane immunofluorescence is still more clearly evidenced on unfixed, suspended erythrocytes (Fig. 2). This anti-blood group A activity of serum 42/f2 can be completely removed by incubating the serum 5 times with an equal volume of washed blood group A1 red cells. The absorbed serum no longer reacts with erythrocytes or erythroblasts, but maintained reactivity with purified CEA (immunodiffusion), colonic tumors and noncancerous colon (immunofluorescence). Sections of spleen and lymph nodes are not stained by serum 42/f2. Anti-NCA sera Double immunodiffusion shows that antiNCA, anti-CEX and anti-NGP sera, and unabsorbed serum 42/f have a common reactivity against perchloric extract of normal human spleen (Fig. 3, 4). O n immunofluorescence, these sera stain the membrane of colo-rectal cancer cells and non-cancerous colon epithelial cells, as previously reported by Burtin et al. [8]. However non-cancerous, round cells, infiltrating the stroma of the tumors, are also strongly positive: brightgreen granular fluorescence is localized to the cytoplasm bringing out, in negative, the nucleus which is polylobed or kidney-shaped (Fig. 5). On blood smears, erythrocytes, platelets and lymphocytes are negative, but the sera stain the cytoplasm of polymorphs, and probably, also, of monocytes (Fig. 4). In bone marrow, normoblasts and megakaryocytes are negative, but the cytoplasm of the cells of the granulopoietic line is stained (Fig. 7) Although. the accurate identification of the cells is difficult, staining seems to increase with the
CEA and NCA in hematopoietic tissues
degree of maturation: polymorphonuclear cells and metamyelocytes are strongly positive, the reactivity of myelocytes and promyelocytes is weaker, and myeloblasts are probably negative. On spleen and lymph node sections, anti-NCA and related antisera react with round cells, usually mononuclear, infiltrating the lymph node sinusoids, marginal zone and red pulp of the spleen, but not in germinal centers (Fig. 8). The positive cells have the shape and location of macrophagic reticulo-endothelial cells. Similar positive cells can be found in the lung, invading the alveolar lumen and the septa. The peritoneal macrophages cultivated from ascitic fluid are also stained by anti-NCA antisera. Absroption experiments demonstrate the common immunologic reactivity of antigen present in cancerous or non-cancerous colonic epithelial cells, and macrophagic or polymorphonuclear cells. After absorption with perchloric extract of normal spleen, or burly coat of a patient with chronic myeloid leukemia, or purified NGP, anti-NCA and related sera no longer stain intestinal carcinoma cells, or epithelial cells of non-cancerous colon. Polymorphonuclear and macrophagic cells are not stained when normal rabbit serum is substituted for anti-NCA serum on ethanol-fixed smears or tissue sections. On the other hand, a bright membrane immunofluorescence was observed when living polymorphs, in suspension, were treated by normal rabbit serum; this non-specific reaction prevents the use of this method for NCA location on the membrane of living, polymorphonuclear cells or macrophages. DISCUSSION
The presence of anti-blood group A reactivity in anti-CEA serum before absorption with blood group A erythrocytes is not surprising, since the purified CEA samples used to immunize rabbits have a blood group A antigenic activity. From the work of Holburn et al. [13], this activity seems to be due to antigenic determinants on the CEA molecule, rather than contamination by blood group A glycoprotein. After absorption of anti blood group A antibodies, anti-CEA serum does not react with blood cells or hematopoietic tissues. This lack of reactivity of anti-CEA serum with erythrocytes is at variance with data of Nery et al. [4], who report partial cross-reactivity between CEA and a material extracted from red blood cell membrane. Disagreement between Nery's and present data could be explained in various ways. The CEA-like antigen could not be recognized by anti-CEA serum used in this work, or could have been buried
785
too deeply in the erythrocyte membrane. On the other hand, membrane immunofluorescence could be quenched by hemoglobin. Antisera directed against the antigen crossreacting with CEA (called NCA, NGP or CEX) strongly stain the cytoplasm of polymorphonuclear cells, myelocytes and mononuclear cells, which probably are macrophages. In addition to this localization, NCA is present in cancerous and non-cancerous epithelial cells of the digestive tract. Staining of the digestive epithelial cells could be related to a residual anti-CEA reactivity of the antisera against CEX, NCA or NGP used in this work. This seems unlikely as these sera stain the adjacent cellular membranes, a localization previously reported by Burtin et al. [8] as specific for NCA, whereas anti-CEA serum only stains the apical pole of the cancerous cells. Furthermore, the absorption experiments performed in this work demonstrate that polymorphonuclear NCA and epithelial NCA have a common immunologic reactivity. Localization in leukocytes of an apparently tumor-associated antigen has been reported previously by Dufour et al [14] in a 2-acetylaminofluorene-induced rat hepatoma. Similar findings were made with dimethylhydrazineinduced rat colonic carcinoma [15]. In both cases rabbit antisera raised against tumor extracts, and made specific by absorption with normal tissues extracts, did not stain tumor cells, but the polymorphonuclear cells present in the tumor, hematopoietic tissues, and normal blood. However, contrary to the abovementioned rat antigen(s), NCA is also present in normal and cancerous epithelial cells. This advances the possibility that the antigen, detected in macrophages and polymorphonuclear cells, could be absorbed by pinocytosis, instead of being locally synthesized. While this hypothesis could be true with regard to the tumour, it can not explain the localization of NCA in myelocytes and metamyelocytes in the bone marrow of patients without cancer. The immunofluorescence staining of colon carcinoma by anti-NCA serum demonstrates the marked infiltration of the tumor stroma by polymorphonuclear and macrophagic cells. Whether these cells are only clearing necrotic tissues, or playing an active role in the host defense against cancer is yet to be determined.
Acknowledgements--This work could not have been done without the reagents generously given by Dr. Burtin (Villejuif), Dr. Darcy (London) and Dr. Mach (Lausanne). The authors wish to express their thanks to Pr. Cahanne, Centre G.F. Leclerc, Dijon, for his material help, and Mr. Thoret, for his excellent technical assistance.
786
M. Bordes, S. Knobel and F. Martin REFERENCES 1. 2. 3. 4. 5.
P. GOLD and S. O. FREEDMAN, Specific carcinoembryonic antigens of the human digestive system. J. exp. Med. 122, 467 (1965). F. MARTIN and M. S. MARTIN, Demonstration of antigens related to colonic cancer in the human digestive system. Int. or. Cancer 6, 352 (1970). G. PUSZTASZERIand J. P. MACH, Carcinoembryonic antigen (CEA) in non digestive cancerous and normal tissues. Immunochemistry 10, 197 (1973). R. NERY, H. BULLMANand A. L. BARSOUM,Carcinoembryonic antigens of erythrocyte membranes. Nature (Lond.) 246, 44 (1973). S. VON KLEIST, G. CHAVANEL and P. BURTIN, Identification of a normal antigen that cross-reacts with the carcinoembryonic antigen. Proc. natl. Acad.
Sci. (Wash.) 69, 2492 (1972). 6.
J . P . MACI~and G. PUSZTASZERI,Carcinoembryonic antigen (CEA) : demonstration of a partial identity between CEA and a normal glycoprotein. Immunochemistry 9, 1031 (1972). 7. D . A . DARCY, C. TURBERVILLEand R. JAMES, Immunological study of carcinoembryonic antigen (CEA) and a related glycoprotein. Brit. J. Cancer 28, 147 (1973). 8. P. BURTIN, S. VON KLEIST, M. C. SABINEand M. KINO, Immunohistological localization of carcinoembryonic antigen and non-specific cross-reacting antigen in gastrointestinal normal and tumoral tissues. Cancer Res. 33, 3299 (1973). 9. F. MARTIN and M. S. MARTIN, Radioimmunoassay of carcinoembryonic antigen in extracts of human colon and stomach. Int. J. Cancer 9, 641 (1972). 10. G. SAINTE-M~RIE, Paraffin embedding technique for studies employing immunofluorescence. J. Histochem. Cytochem. 10, 250 (1962). 11. M. BORDES, R. MICHIELSand F. MARTIN, Detection by immunofluorescence of carcinoembryonic antigen in colonic carcinoma, other malignant or benign tumours, and non-cancerous tissues. Digestion 9, 106 (1973). 12. P. BURTIN and M. C. SABINE, Inhibition of the non-specific fixation of fluorescent globulins. Rev. europ. Etud. clin. biol. 17, 76 (1972). 13. A . M . HOLBURN,J. P. MACH, D. MACDONALDand M. NEWLANDS, Studies of the association of the A, B and Lewis blood group antigens with carcinoembryonic antigen (CEA). Immunology 26, 831 (1974). 14. D. DUFOUR, S. LEMIEUX, A. TREMBLAYand R. ESTRADA, Immunochemical studies of extramedullary hematopoiesis and of the participation of leukocytes in carcinogenesis. In Embryonic and Fetal Antigens in Cancer. (Edited by N. G. ANDERSONandJ. H. COGGIN)Vol. 2, p. 199. Oak Ridge National Laboratory, Oak Ridge, Tennessee (1972). 15. F. MARTIN, M. S. MARTIN, M. BORDESand S. KNOBEL,Antigens associated with intestinal carcinomas chemically induced in rats. Int. J. Cancer 15, 144 (1975).
Carcinoembryonic Antigen (CEA)and Related Antigens in Blood Cells and Hematopoietic Tissues* M. BORDES, S. KNOBEL and F. MARTIN Laboratory of Experimental iVIedicine, I N S E R M U-45, Facultd de Mbdecine, D~on, (France). Abstract--Indirect immunofluorescence was used to detect and localize carcinoembryonic antigen (CEA) and non-specific cross-reacting antigen ( NCA) in blood cells and hematopoietic tissues. CEA was neverfound, but NCA could easily be identified in the cytoplasm of blood polymorphonuclear cells, bone marrow myelo~ytes, and monocytic cells (probably macrophages) infiltrating the spleen and lymph nodes. In normal colon and colonic adenocarcinoma, CEA was present in the walls of epithelial cells, and in round, isolated polymorphonuclear or mononuclear cells, infiltrating the submucosa and the stroma of the tumor.
INTRODUCTION
MATERIAL AND METHODS
CARCINOEMBRYONIC antigen (CEA) was first defined as a cancer- and organ-specific antigen, restricted to cancerous and fetal digestive tissues [1]. This specificity was later questioned, and CEA was detected in various epithelial tissues and secretions [2, 3]. CEA, or an immunologically related antigen, has recently been demonstrated on the walls of normal erythrocytes [4]. A glycoprotein, sharing a common antigenic site with CEA, was independently substantiated by several groups, and called non-specific cross-reacting antigen(NCA) [5], normal glycoprotein (NGP) [6] and C E X [7]. Exchange of reagents has proven the immunologic identity of NCA, N G P and C E X [7]. This antigen has been found, in association with CEA, in normal and cancerous epithelial tissues [8], but was also present in normal spleen [5], an organ which does not contain epithelial tissue. The present work takes into consideration the location of CEA and NCA in blood cells and hematopoietic tissues.
Antigens and antisera Purified CEA was prepared in this laboratory, according to previously reported procedure [9]. Anti-CEA serum was prepared by repeated immunization of rabbit with purified CEA. This serum, referred to as 42/f, has both antiCEA and anti-NCA reactivities. It was made specific for CEA by absorption of perchloric acid extract from normal human spleen (50 mg/ml). This absorbed serum will be referred to as 42/f2. Blood and tissue specimens Samples of blood have been obtained from 20 healthy donors of various blood group status. Morphologically normal, sternal bone marrow samplings came from 5 patients with mild anaemia, of non-malignant origin. Smears of blood and bone marrow were immediately fixed for 20 min in 95 % ethanol. The other tissues were fixed and embedded according to Sainte-Marie [10]. The non-cancerous specimens were 2 spleens, surgically removed after a traumatic rupture, 2 cervical lymph nodes (biopsies), with mild subacute lymphadenitis,
Accepted 1 April 1975. *This work was supported by research grants CA-401 and 74.1.080.04 from the Institut National pour la Sant6 et la Recherche Mddicale. 783
Fig. 5. Colonic adenocarcinoma ~tained by anti-NGP antiserum (indirect immunofluorescence). The membrane of tumor cells (bottom, right), but also round cells infiltrating the stroma (arrow) are stained. ( x 550). Fig. 6.
Circulating blood polymorphonuclear cell stained by anti-NGP antiserum (indirect immunofluorescence). ( x 550).
Fig. 7.
A clone of cells of the granulopoietic line stained by anti-NGP antiserum (indirect immunoJtuorescence) on a bone marrow smear. ( x 550).
Fig. 8. Section of normal spleen, stained by anti-NGP antiserum. The lymphocytes of the germinal center are negative, but the round cells (polymorphonuclear and~or macrophages) around the center are stained. ( x 200).
Fig. 1. Reaction of anti-CEA antiserum (42/f2, central well) with purified CEA and perchloric acid extract of colonic carcinoma (CT) ; the serum does not react with perchloric acid extract of normal spleen (Sp). Fig. 2. illembrane immunofluorescence of blood group A erythrocytes with antiCEA serum, before absorption with A red blood cells. This fluorescence is completely extinguished by absorption of the serum with blood group A erythrocytes. ( x 550). Fig. 3. Reaction of anti-CEX antiserum (X, central well) with perchloric acid extracts of normal spleen (Sp), and polymorphonuclear cells (Pm) and purified NGP. Fig. 4. Reaction of an anti-CEA antiserum non absorbed for NCA (42f, central well) with purified CEA, perchloric acid extracts of colonic carcinoma (CT) and normal spleen (Sp).
(to face p. 784)
784
M. Bordes, S. Knobel and F. Martin
and a morphologically normal lung obtained at autopsy. Three colonic adenocarcinomas, and their adjacent, non-cancerous mucosa (surgical material), were also used in this work. Peritoneal macrophages These were prepared from the ascitic fluid of a patient with alcoholic liver cirrhosis. Five ml were disseminated into tissue culture flasks (Falcon), poured off after 2 hr, and replaced by Ham's FIO medium, supplemented by 20 % fetal calf serum. The adhering cells were mostly macrophages, as proven by phagocytosis of iron carbonyl. After 3 days of culture, they were fixed by absolute ethanol and tested for immunofluorescence. Immunofluorescence staining The indirect immunofluorescence technique, used to stain smears and sections, has been previously reported [11]. 42/f, 42/f2, anti-CEX, anti-NGP and anti-NCA sera were respectively diluted as follows 1/80, 1/80, 1/20, 1/20 and 1/100 according to preliminary results. In the second step, the fluorescein isothiocyanateconjugated reagents were anti-goat globulin rabbit antiserum (Institut Pasteur, Paris) diluted 1/10 for anti-CEX serum and antirabbit globulin sheep antiserum (Institut Pasteur, Paris) diluted 1/10, for the other sera. Special precautions were taken to avoid nonspecific reactions. All the smears and sections were treated by sodium nitrite-acetic acid in accordance with Burtin and Sabine [12]. Fluorescein labeled sera were absorbed by acetonic extracts of normal human spleen (150 mg/ml). Normal rabbit serum was systematically used as a negative control. Observations were made with a Leitz Orthoplan microscope with a vertical illuminator. Membrane immunofluorescence Erythrocytes from blood collected on sodium oxalate were washed 3 times in phosphatebuffered saline (PBS), incubated for 20 rain at room temperature with an equal volume of antiserum, diluted as mentioned above. The red blood cells were washed 3 times in PBS, then incubated in the same volume of fluorescein-labelled serum diluted 1/10. After 3 further washings, they were mounted on slides and examined for membrane immunofiuorescence Immunodiffusion Double immunodiffusion was carried out according to a modification of Ouchterlony's technique [2].
RESULTS
Normal rabbit serum On immunofluorescence, normal rabbit serum, used as control, does not stain ethanolfixed smears or tissue sections of hematopoietic tissues or colonic adenocarcinoma. Anti-CEA serum On immunodiffusion (Fig. 1), anti-CEA serum 42/f 2 gives a precipitin line with perchloric extract of colonic tumor or purified CEA, but does not react with perchloric extracts of spleen. The immunofluorescence staining of colonic tumors and non cancerous colonic mucosa by this serum has been described elsewhere [11]. When tested on fixed blood or bone marrow smears, serum 42/f2 stains the membrane of erythrocytes and erythroblasts from patients of blood group A or AB, but not of O or B. Membrane immunofluorescence is still more clearly evidenced on unfixed, suspended erythrocytes (Fig. 2). This anti-blood group A activity of serum 42/f2 can be completely removed by incubating the serum 5 times with an equal volume of washed blood group A1 red cells. The absorbed serum no longer reacts with erythrocytes or erythroblasts, but maintained reactivity with purified CEA (immunodiffusion), colonic tumors and noncancerous colon (immunofluorescence). Sections of spleen and lymph nodes are not stained by serum 42/f2. Anti-NCA sera Double immunodiffusion shows that antiNCA, anti-CEX and anti-NGP sera, and unabsorbed serum 42/f have a common reactivity against perchloric extract of normal human spleen (Fig. 3, 4). O n immunofluorescence, these sera stain the membrane of colo-rectal cancer cells and non-cancerous colon epithelial cells, as previously reported by Burtin et al. [8]. However non-cancerous, round cells, infiltrating the stroma of the tumors, are also strongly positive: brightgreen granular fluorescence is localized to the cytoplasm bringing out, in negative, the nucleus which is polylobed or kidney-shaped (Fig. 5). On blood smears, erythrocytes, platelets and lymphocytes are negative, but the sera stain the cytoplasm of polymorphs, and probably, also, of monocytes (Fig. 4). In bone marrow, normoblasts and megakaryocytes are negative, but the cytoplasm of the cells of the granulopoietic line is stained (Fig. 7) Although. the accurate identification of the cells is difficult, staining seems to increase with the
CEA and NCA in hematopoietic tissues
degree of maturation: polymorphonuclear cells and metamyelocytes are strongly positive, the reactivity of myelocytes and promyelocytes is weaker, and myeloblasts are probably negative. On spleen and lymph node sections, anti-NCA and related antisera react with round cells, usually mononuclear, infiltrating the lymph node sinusoids, marginal zone and red pulp of the spleen, but not in germinal centers (Fig. 8). The positive cells have the shape and location of macrophagic reticulo-endothelial cells. Similar positive cells can be found in the lung, invading the alveolar lumen and the septa. The peritoneal macrophages cultivated from ascitic fluid are also stained by anti-NCA antisera. Absroption experiments demonstrate the common immunologic reactivity of antigen present in cancerous or non-cancerous colonic epithelial cells, and macrophagic or polymorphonuclear cells. After absorption with perchloric extract of normal spleen, or burly coat of a patient with chronic myeloid leukemia, or purified NGP, anti-NCA and related sera no longer stain intestinal carcinoma cells, or epithelial cells of non-cancerous colon. Polymorphonuclear and macrophagic cells are not stained when normal rabbit serum is substituted for anti-NCA serum on ethanol-fixed smears or tissue sections. On the other hand, a bright membrane immunofluorescence was observed when living polymorphs, in suspension, were treated by normal rabbit serum; this non-specific reaction prevents the use of this method for NCA location on the membrane of living, polymorphonuclear cells or macrophages. DISCUSSION
The presence of anti-blood group A reactivity in anti-CEA serum before absorption with blood group A erythrocytes is not surprising, since the purified CEA samples used to immunize rabbits have a blood group A antigenic activity. From the work of Holburn et al. [13], this activity seems to be due to antigenic determinants on the CEA molecule, rather than contamination by blood group A glycoprotein. After absorption of anti blood group A antibodies, anti-CEA serum does not react with blood cells or hematopoietic tissues. This lack of reactivity of anti-CEA serum with erythrocytes is at variance with data of Nery et al. [4], who report partial cross-reactivity between CEA and a material extracted from red blood cell membrane. Disagreement between Nery's and present data could be explained in various ways. The CEA-like antigen could not be recognized by anti-CEA serum used in this work, or could have been buried
785
too deeply in the erythrocyte membrane. On the other hand, membrane immunofluorescence could be quenched by hemoglobin. Antisera directed against the antigen crossreacting with CEA (called NCA, NGP or CEX) strongly stain the cytoplasm of polymorphonuclear cells, myelocytes and mononuclear cells, which probably are macrophages. In addition to this localization, NCA is present in cancerous and non-cancerous epithelial cells of the digestive tract. Staining of the digestive epithelial cells could be related to a residual anti-CEA reactivity of the antisera against CEX, NCA or NGP used in this work. This seems unlikely as these sera stain the adjacent cellular membranes, a localization previously reported by Burtin et al. [8] as specific for NCA, whereas anti-CEA serum only stains the apical pole of the cancerous cells. Furthermore, the absorption experiments performed in this work demonstrate that polymorphonuclear NCA and epithelial NCA have a common immunologic reactivity. Localization in leukocytes of an apparently tumor-associated antigen has been reported previously by Dufour et al [14] in a 2-acetylaminofluorene-induced rat hepatoma. Similar findings were made with dimethylhydrazineinduced rat colonic carcinoma [15]. In both cases rabbit antisera raised against tumor extracts, and made specific by absorption with normal tissues extracts, did not stain tumor cells, but the polymorphonuclear cells present in the tumor, hematopoietic tissues, and normal blood. However, contrary to the abovementioned rat antigen(s), NCA is also present in normal and cancerous epithelial cells. This advances the possibility that the antigen, detected in macrophages and polymorphonuclear cells, could be absorbed by pinocytosis, instead of being locally synthesized. While this hypothesis could be true with regard to the tumour, it can not explain the localization of NCA in myelocytes and metamyelocytes in the bone marrow of patients without cancer. The immunofluorescence staining of colon carcinoma by anti-NCA serum demonstrates the marked infiltration of the tumor stroma by polymorphonuclear and macrophagic cells. Whether these cells are only clearing necrotic tissues, or playing an active role in the host defense against cancer is yet to be determined.
Acknowledgements--This work could not have been done without the reagents generously given by Dr. Burtin (Villejuif), Dr. Darcy (London) and Dr. Mach (Lausanne). The authors wish to express their thanks to Pr. Cahanne, Centre G.F. Leclerc, Dijon, for his material help, and Mr. Thoret, for his excellent technical assistance.
786
M. Bordes, S. Knobel and F. Martin REFERENCES 1. 2. 3. 4. 5.
P. GOLD and S. O. FREEDMAN, Specific carcinoembryonic antigens of the human digestive system. J. exp. Med. 122, 467 (1965). F. MARTIN and M. S. MARTIN, Demonstration of antigens related to colonic cancer in the human digestive system. Int. or. Cancer 6, 352 (1970). G. PUSZTASZERIand J. P. MACH, Carcinoembryonic antigen (CEA) in non digestive cancerous and normal tissues. Immunochemistry 10, 197 (1973). R. NERY, H. BULLMANand A. L. BARSOUM,Carcinoembryonic antigens of erythrocyte membranes. Nature (Lond.) 246, 44 (1973). S. VON KLEIST, G. CHAVANEL and P. BURTIN, Identification of a normal antigen that cross-reacts with the carcinoembryonic antigen. Proc. natl. Acad.
Sci. (Wash.) 69, 2492 (1972). 6.
J . P . MACI~and G. PUSZTASZERI,Carcinoembryonic antigen (CEA) : demonstration of a partial identity between CEA and a normal glycoprotein. Immunochemistry 9, 1031 (1972). 7. D . A . DARCY, C. TURBERVILLEand R. JAMES, Immunological study of carcinoembryonic antigen (CEA) and a related glycoprotein. Brit. J. Cancer 28, 147 (1973). 8. P. BURTIN, S. VON KLEIST, M. C. SABINEand M. KINO, Immunohistological localization of carcinoembryonic antigen and non-specific cross-reacting antigen in gastrointestinal normal and tumoral tissues. Cancer Res. 33, 3299 (1973). 9. F. MARTIN and M. S. MARTIN, Radioimmunoassay of carcinoembryonic antigen in extracts of human colon and stomach. Int. J. Cancer 9, 641 (1972). 10. G. SAINTE-M~RIE, Paraffin embedding technique for studies employing immunofluorescence. J. Histochem. Cytochem. 10, 250 (1962). 11. M. BORDES, R. MICHIELSand F. MARTIN, Detection by immunofluorescence of carcinoembryonic antigen in colonic carcinoma, other malignant or benign tumours, and non-cancerous tissues. Digestion 9, 106 (1973). 12. P. BURTIN and M. C. SABINE, Inhibition of the non-specific fixation of fluorescent globulins. Rev. europ. Etud. clin. biol. 17, 76 (1972). 13. A . M . HOLBURN,J. P. MACH, D. MACDONALDand M. NEWLANDS, Studies of the association of the A, B and Lewis blood group antigens with carcinoembryonic antigen (CEA). Immunology 26, 831 (1974). 14. D. DUFOUR, S. LEMIEUX, A. TREMBLAYand R. ESTRADA, Immunochemical studies of extramedullary hematopoiesis and of the participation of leukocytes in carcinogenesis. In Embryonic and Fetal Antigens in Cancer. (Edited by N. G. ANDERSONandJ. H. COGGIN)Vol. 2, p. 199. Oak Ridge National Laboratory, Oak Ridge, Tennessee (1972). 15. F. MARTIN, M. S. MARTIN, M. BORDESand S. KNOBEL,Antigens associated with intestinal carcinomas chemically induced in rats. Int. J. Cancer 15, 144 (1975).
